Home |  Company Profile |  Contacts |  Access Map |  Technical Articles |  Literature |  Version Française de Cette Page


In recent years the hybrid microelectronics industry has been plagued with an apparent phenomenon or condition called "one-way leaker" on various styles of hermetic glass-to-metal seal packages. The application of monolithic integrated circuit requirements for hermeticity testing of large area hybrid devices, Mil-Std-883, Test Method 1014, Condition C further highlights the problem. The condition appears to be most prevalent under the 60 psig gross leak bomb pressure and least prevalent under 30 psig bomb pressure.
The fine and gross leak testing of Mil-Std-883, Test Method 1014, subject the sealed hybrid packages to various combinations of time-pressure-temperature stresses. Under various bomb pressures, the intergranular oxide of the glass-to-metal seat appears to be temporarily stressed, thereby developing minute cracks.
During the bombing process, the leak test fluids, along with other gases and potential contaminants, can be injected into the sealed packages. After removal of the various pressures, the developed microcracks apparently reseal, become hermetic again, and test good. With the possible exception of the Radioisotope Test Method, Test Method 1014, Test Condition B, current standard detection methods, including the weight-gain measurement, are incapable of detecting these one way leak conditions.
The term "one-way leaker" is sometimes used to describe a hybrid which passes all Mil-Std leak requirements yet still provides a Residual Gas Analysis (RGA) spectrum containing large amounts of air. The controversial assumption generally made is that the part was hermetic under one set of conditions and 'leaky' under a different set of conditions. The controversy stems from the fact that the most common method of forming a glass to metal seal in Kovar packages is by means of oxide bonding. Such seals, once broken, should not be expected to reseal themselves. This paper details the results of a failure analysis of an RGA failure with a power hybrid case where the bow in the base of the package was found to have resulted in sufficient stress during burn-in to create a 'one-way' leaker. The leak location was subsequently confirmed by dye penetrant testing. A conclusive experimental verification of how the leak rate of such a glass to metal seal can vary with pressure and/or temperature is presented.
By the late '80's various microelectronic device manufacturers had experienced the "Hydrogen Phenomenon" knowingly and unknowingly. This phenomenon occurs when residual or absorbed hydrogen has remained within microstructure trap sites of the ferrous alloy packaging materials that they had selected. But as a function of burn-in or other thermal stresses, the hydrogen is desorbed into the cavity of the device. A variety of chemical reactions are then potentially available by which the desorbing hydrogen gas can destroy device integrity and product reliability. To date, a multitude of subsequent chemical reactions have been identified which can be cause for the failure of the product. This paper identifies an in depth analytical routine which has allowed the supplier community to provide the user community with hydrogen free packages. Potential sources or traps for absorbed hydrogen have been theoretically identified. "Bake out" procedures, and the affects of annealing and plating are also reviewed, all of which have been found to ultimately impact device reliability, i.e., should product design, and process variables all align contrarily to the needs of the product type in question.
The phenomenon of Hydrogen Desorption was first observed on captive microwave devices made with Gallium Arsenide die. Titanium adhesion metallurgy also used as in-line resistors were noted to become bumpy and even lose adhesion. Meanwhile the circuits were noted to electrically drift. Residual Gas Analysis (RGA) of this device showed increasing concentrations of hydrogen as a result of thermal stress, e.g., burn-in.
Other device technologies were noted to acquire increasing concentrations of moisture such that in several cases Quality Conformance Inspection (QCI) criteria of Mil-Std-888 Test Method 5008 could not be met. Concurrent with the increase in moisture was the noted increase in hydrogen concentrations. In some cases residual traces of normal air were also noted to change as Argon concentrations remained low, e.g., 100 ppm, but the expected oxygen level at circa 2000 ppm was totally absent.
These incidents all drew their origin from the absorbed hydrogen slowly desorbing into the cavity of the respective devices. Hydride formation with metal systems, such as titanium, and metal oxide reduction, such as the reduction of silver solder glass die attach materials, are now understood to be the root cause for these problems.
IBM personnel set in place a team of industrial participants to help in clarifying the problem. A base metal supplier, packaging houses and an independent analytical facility participated in the study which is detailed in the following paragraphs.
The potentially detrimental effects of outgassed hydrogen on device reliability have been substantiated in the trade press over the past several years. Non-specific to silicon or gallium arsenide technology, its negative impact is universal and can range from hydride formation with ensuing material deformation to oxide reduction followed by moisture-related failure mechanisms. In keeping with the microelectronics industry's move toward "building-in reliability," this article focuses on identifying one of the sources of outgassed hydrogen in hermetically sealed devices and offers some prospective alternatives to processing and assembly to aid in its prevention.
Moisture ingress in electronic packages can lead to catastrophic failures due to electromigration and corrosion. For space application, epoxy sealed CCDs are often used, and the risk due to moisture ingress during test and storage rarely assessed. This article propose a methodology to quantify the moisture ingress speed through a sealed joint and the evolution of the moisture amount inside the cavity.
In a first step, a characterization of the organic materials is carried out. The diffusion and saturation coefficient of the moisture inside the material are calculated. Then, a 2D finite element model is built using Fick diffusion laws, and taking into account the seal, the gas and each polymeric material inside the cavity.
In the next step, the moisture concentration inside the cavity of the CCD package is monitored by means of humidity sensors throughout the experiment. Changing the moisture level of the atmosphere surrounding the package bring changes to the internal moisture content. Both ingress and release of moisture have been observed over several months. The comparison between empty and fully equipped cavities showed the influence of the various materials used inside the cavity on measurable moisture.
Finally, the experimental results are correlated with the model.
The microelectronics community has long been plagued with the problem of moisture formation and out-gassing of various fixed and organic gaseous species into the device cavity. It is now very apparent that the optoelectronic packaging community is having the same problems only made more complex by the use of inadequate test methods, unproven materials and misconceptions in the supply and user industries. A test protocol is provided that addresses these issues, which allows the optoelectronic community to improve device quality and reliability.
RL/NIST Workshop On Moisture measurement and Control for Microelectronics
The workshop, fifth in a series concerned with measurement problems in integrated circuit processing and assembly, served as a forum to examine present problems with the measurement of moisture in hermetic semiconductor devices. While moisture-induced failure modes and mechanisms had received considerable attention in the published literature and in meetings, the accurate and reliable measurement of the moisture content had not; yet, this lack of measurement of moisture reliability is a major obstacle to meaningful efforts to limit and control this pervasize contaminant. Manuscripts and summaries are provided of 19 talks, panel meetings, and group encounters on three major topics: mass spectrometer measurements of internal package moisture, moisture sensors, and package analysis and quality assurance.
The Workshop, one of a series concerned with measurement problems in integrated circuit processing and assembly, served as a forum to examine the progress that has been made in the measurement and control of moisture in hermetically packaged semiconductor devices. While moisture-induced failure modes and mechanisms have been extensively documented, the lack of accurate and reliable measurement of the moisture content itself has been a major obstacle to meaningful efforts to limit and control this pervasive contaminant. Manuscripts are provided of 36 presentations which detail the progress that has been made in mass spectrometer measurements and calibration of internal package moisture, in increased assurance with moisture sensors, in testing, and in package control.
The workshop, one of a series concerned with measurement problems in integrated circuit processing and assembly, served as a forum to examine the continuing progress that has been made in the measurement and control of moisture in hermetically packaged semiconductor devices. Thirty-four presentations are included which contain detailed information for securing hermetic packages with low moisture content. Agreement in measurement has been obtained with the mass spectrometer for cerdip and metal packages at the 5000 ppmv level of moisture through the use of suitable moisture generators, a 3-volume calibrator, calibrated dewpoint hygrometers, and appropriate operational procedures. An approach is given for a reproducible and reliable transfer package. However, the increased use of organic materials in new and rapidly expanding technologies such as VLSI/VHSIC and hybrid packaging presents new and more complex challenges to accurate measurement of interior moisture.
This fourth Workshop on Moisture Measurement and Control for Microelectronics served as a forum on moisture and/or materials reliability problems and on ways to control them or measure their extent. Twenty-two presentations are included which contain detailed information on hermeticity measurement and definition; development of standard packages for mass spectrometric calibrations; moisture interaction with various materials; and techniques that can be used to measure moisture microelectronics. It was clear from several presentations in this workshop that a very systematic approach is needed when organic materials are involved; all the variables must be identified and studied one at a time. This is the key to lot-to-lot reproducibility, materials selection, and control; hence a better reliability at the design phase will decrease the need for testing, hence the cost, thus resulting in a greater satisfaction to the customer.
Books of Interest
The technology of hermeticity addresses the transfer of fluids in and out of sealed enclosures. This technology is based on physics and chemistry, and (like many such technologies) is difficult to grasp when the exposure is brief or infrequent. One's first exposure to this technology usually involves an application related problem. The understanding of, and particularly the solution to, the problem requires a considerable specific background. Not having such a background, the physical concept of the problem is just out of one's grasp and its solution is nowhere in sight. Subsequent exposure to this technology only helps a little, as the background is still missing and the new application is often slightly different.
The purpose of this monograph is to provide the necessary background and problem solving examples, so that packaging engineers and other specialists can apply this knowledge to solving their problems. Ninety nine problems and their solutions are presented. These problems are representative of the type of problems occurring in industry. Many of the included problems are those that the author has experienced.
The technology of hermeticity is an offshoot from vacuum science. Vacuum science has a long history, going back to two Italians: Gasparo Berti in 1640, and Evangelista Torricelli in 1644. During the next three hundred and some years, scientists have tried to produce better and better vacuums. They realized that the degree of vacuum achieved, not only depends upon how much and how fast the gas can be removed from the vessel, but also upon the amount and rate of gas leaking into the vessel. This lack of an hermetic vessel eventually led to the technology of hermeticity.
One method of finding leaks in a vacuum system was to connect the system to a mass spectrometer which was tuned to the gas; helium. Helium was selected because the amount of helium in the atmosphere is only 1 part in 200,000 (the rate of its diffusion through a leak is greater than any other gas except hydrogen), and that no other gas can be mistaken for helium by a mass spectrometer. Helium was then sprayed at various parts of the system and if there was a leak, the mass spectrometer would so indicate. This technique, slightly modified, would eventually be used to detect leaks in sealed packages when they contained helium.
The leak testing of sealed packages, when the initial atmosphere in the enclosure had some helium, became a common practice by the early nineteen sixties. In 1965 D. A. Howl and C. A. Mann reported on a leak testing method for enclosures which were not sealed in an atmosphere containing helium. This new method forced helium under pressure through the leakage path into the enclosure. A helium mass spectrometer then detected the helium escaping the enclosure. Subsequently, MIL-STD 883 adopted a leak test method based on this work.
Bibliographies at the end of chapters will lead the reader to areas beyond the present scope of this monograph.